|Home | About | Journals | Submit | Contact Us | Français|
A framework for understanding and evaluating physicians' skills at providing end of life care from the perspectives of patients, families, and health care workers will promote better quality of care at the end of life.
To develop a comprehensive understanding of the factors contributing to the quality of physicians' care for dying patients.
Qualitative study using focus groups and content analysis based on grounded theory.
Eleven focus groups of patients with chronic obstructive pulmonary disease, AIDS, or cancer (79 patients); 3 groups of family members who had a loved one die of chronic disease (20 family members); 4 groups of nurses and social workers from hospice or acute care settings (27 health care workers); and 2 groups of physicians with expertise in end-of-life care (11 physicians).
We identified 12 domains of physicians' skills at providing end-of-life care: accessibility and continuity; team coordination and communication; communication with patients; patient education; inclusion and recognition of the family; competence; pain and symptom management; emotional support, personalization; attention to patient values; respect and humility; and support of patient decision making. within these domains, we identified 55 specific components of physicians' skills. Domains identified most frequently by patients and families were emotional support and communication with patients. Patients with the 3 disease groups, families, and health care workers identified all 12 domains. Investigators used transcript analyses to construct a conceptual model of physicians' skills at providing end-of-life care that grouped domains into 5 categories.
The 12 domains encompass the major aspects of physicians' skills at providing high-quality end-of-life care from the perspectives of patients, their families, and health care workers, and provide a new framework for understanding, evaluating, and teaching these skills. Our findings should focus physicians, physician-educators, and researchers on communication, emotional support, and accessibility to improve the quality of end-of-life care.
Increasing attention has focused on the quality of end-of-life care in the medical literature1 and the lay press.2 A report from the Institute of Medicine recently concluded, “Many are denied the chance to die well because of inadequate medical care.”3 It appears that the quality of end-of-life care provided by physicians is poor: a large number of hospitalized patients are dying in pain,4 patients are referred to hospice late in the course of their terminal illness,5 and many physicians report acting against their consciences in providing end-of-life care.6 Randomized controlled trials of interventions intended to improve care at the end-of-life have not shown significant improvements in care.4,7,8 One of the reasons investigators have been unsuccessful in demonstrating improvements in end-of-life care may be that we do not have a complete understanding of the components of the quality of this care that are important to patients, their families, and health care workers.9 Patients with chronic or terminal diseases and their families represent an important perspective because they are the recipients of this care. Qualitative research offers the opportunity to elicit and understand the aspects of physician skill at providing end-of-life care that patients and families deem most important.
Experts have published opinions concerning the components of quality of end-of-life care3,10,11 and one prior qualitative study examined patients' perspectives on quality of end-of-life care.12 Although these reports have previously identified some general domains, no one has used qualitative methods to identify, from all major stakeholders, the specific components of excellent quality end-of-life care provided by physicians. The objective of this study was to develop an integrated taxonomy of the major domains and specific components of physicians' skills at providing excellent quality end-of-life care from the perspectives of patients with terminal illnesses, family members who have lost loved ones to chronic diseases, and health care workers with experience providing end-of-life care.
In this qualitative study, we used focus groups to determine the perspectives of terminally ill patients, family members, and health care professionals concerning physicians' skills at providing end-of-life care. Focus groups are a qualitative study method that capitalizes on group dynamics to obtain information that may not be available through individual interviews.13,14 The University of Washington Human Subjects Committee approved all procedures.
We conducted 20 focus groups formed on unifying features. To elicit the perspectives of patients, 3 groups were held with patients with chronic obstructive pulmonary disease (COPD) (n = 24), 4 groups with patients with AIDS (n = 36), and 4 groups with patients with cancer (n = 19.) We also conducted 3 groups consisting of family members who had lost a loved one to a chronic disease within the preceding 2 years (n = 20), 4 groups consisting of nurses or social workers from hospice or acute care settings (n = 27), and 2 groups of practicing physicians with expertise in end-of-life care (n = 11).
The goal of the purposive sampling used in this study was to identify individuals who were interested in discussing and knowledgeable about physicians' skills at providing end-of-life care. We used convenience sampling to identify participants until we had a sufficient number for at least 3 focus groups in each category (we used only 2 groups of physicians because the investigative group and consulting experts also contained physicians). In recruiting patients, we described the study focus as developing an understanding of physicians' skills at providing care to patients with chronic and life-threatening diseases in order to avoid excluding patients who did not view themselves as near the “end of life.” Patients were recruited by posting flyers in academic and community-based physicians' offices and through support groups for patients with COPD, AIDS, or cancer. Inclusion criteria for patients were as follows: oxygen-dependent COPD, C3 AIDS as defined by the Centers for Disease Control and Prevention,15 or metastatic cancer or nonoperable lung cancer. Family members were recruited through flyers, grief support groups, and key informants in hospice programs. Nurses and social workers were recruited through key informants in university and community intensive care units or acute medical wards or through coordinators at hospice programs. Physicians were identified by asking participating nurses to name 3 outstanding physicians who provided end-of-life care.
Focus groups were conducted by a trained qualitative researcher (DMA) with a moderator's guide that included 4 scripted questions. Questions for patient groups were as follows:
In addition, the focus group leader used scripted probes, when needed, to continue discussion of a specific question or to keep the focus on end-of-life care. Questions used in family and health care provider groups were similar (moderator guides available from the authors). The moderator was responsible for encouraging all participants to voice opinions and preventing individuals from monopolizing discussion. Focus groups were 90 minutes in length, and participants were paid $30. All focus groups were audiotaped and transcribed verbatim.
All transcripts were analyzed using principles of grounded theory.16,17 Domains of physicians' skills at providing end-of-life care were identified using open coding. Specific components within each domain were then identified using selective coding, and a conceptual model was developed using axial coding. Explanations are provided below for these 3 types of coding.
Investigators used previously published domains of quality of end-of-life care to draft a preliminary framework for the domains.3,11,18 This preliminary framework was drafted during 2 full-day meetings of investigators with input from 3 additional experts in medical ethics or end-of-life care from the University of Washington (Albert R. Jonsen, PhD; Anthony L. Back, MD; Stuart J. Farber, MD) and 2 experts from other institutions (Susan D. Block, MD, Harvard University; Joan M. Teno, MD, MSc, Brown University).
To revise the preliminary framework and develop the initial domain list for coding transcripts, each investigator independently reviewed the same 4 focus group transcripts, coding all relevant passages of participant speech into one or more of the preliminary domains. Investigators met to review codes and revise the preliminary domain list. In this process, one of the preliminary domains, ethical and legal decision making, was eliminated, and several new domains were developed, including personalization, accessibility, continuity, and patient education. After the framework was revised, investigators formed 3 pairs, and each pair reviewed 6 or 7 transcripts; all 20 transcripts were reviewed and coded. The investigators independently coded all relevant passages of participant speech (defined as the entire segment of speech until a change in speakers) into 1 or more of the revised domains, and then reviewed coding with the other investigator in the pair. Discrepancies were discussed within a pair, and if resolution was not achieved, the passage was brought to the entire group of investigators for discussion. During this process, the domain structure was continually reassessed and underwent subsequent revisions (for example, accessibility and continuity were combined into one domain).
We used qualitative software (Atlas-TI, Scientific Software Development, Thousand Oaks, Calif.) to identify all passages coded for each domain. Teams of 2 or 3 investigators then reviewed all passages within a domain to identify the major themes or “components” of the domain (selective coding). The investigators identified a minimum of three passages that captured the component and used the words of patients or families to label the component. The entire group of investigators reviewed all components and representative passages to ensure veracity of the components and wording. The language of patients and families was used to label components (and is used to illustrate domains in this paper) because we believe it is this perspective that is most in need of being captured to improve end-of-life care. However, the domains and components also reflect the perspective of health care workers.
After development of domains and components, investigators held a series of meetings to develop a conceptual model. We used the information and context obtained from review of the transcripts to identify connections between domains (axial coding) and place domains into a model.
There are valid alternative ways of viewing the placement of components into domains and domains into the conceptual model. Some alternatives have been suggested during presentation of this material and in the editorial process. We have taken these suggestions into consideration along the way, and made the decision to resist making changes based on the input of observers outside the qualitative analysis process described above.
We used 2 methods to assess the trustworthiness of the coding and data. First, to assess the reliability of domain coding, each investigator-pair reviewed and coded the same 3 transcripts using the methods described above. We compared coding across the 3 investigator-pairs and found agreement for 63% of codes across all three pairs and agreement for 89% of codes for 2 of 3 pairs.
To help assess the content validity of the components, we devised a written questionnaire using all components. We identified 30 patients (10 with COPD, 10 with AIDS, and 10 with cancer) from academic and community-based physician practices and 20 family members of dying or deceased patients for one-on-one interviews. None of these individuals had participated in the focus groups. Participants were asked to complete a questionnaire in which they rated a specific physician involved in their care (patients) or their loved one's care (family members) for each component. Participants were asked to identify any items that were unclear or worded in a way that was bothersome and discussed these issues in a cognitive debriefing interview conducted by a research assistant. Finally, we also identified 20 health care workers (nurses or social workers) and 20 practicing physicians to rate a physician colleague with whom they had worked in end-of-life care using the questionnaire and assess the clarity of the components. During this process, no component was identified as being unclear by more than 10% of the participants and the participants endorsed the accuracy and comprehensiveness of the components.
Table 1 shows the number of participants in each category and the demographic characteristics of focus group participants. As shown in Table 2, investigators identified 12 domains and 55 specific components of physicians' skills at providing high quality end-of-life care from the focus group transcripts. Each domain was identified in every group, suggesting that additional groups would be unlikely to yield additional domains. Table 3 shows the proportion of coded passages that fell within each domain for the different types of participants and the rankings for the frequency with which each domain was identified.
Communication with Patients. Communication with patients was consistently 1 of the 3 domains most frequently identified by all groups, and includes items concerning the ability of the physician to listen, invite questions, speak honestly and sensitively with patients, and talk about dying and death. Examples of quotes from this domain include:
Like when I had my first PCP outbreak, I didn't want to have a certain thing done to me because it terrified me. And I had a lot of doctors yelling at me, “You have to do this, you have to do that,” but then there was one who really listened to me… he was just there to observe and he was the only one who listened to me.
(Patient with AIDS)
My experience… is that doctors don't really like to talk about death… or know how to talk about death. And that was very difficult for us, because this was our first experience with a family member dying… Every time we got the truth, mostly from the hospice nurses, we really appreciated that… it focused us on the issues so we could have some quality time with my dad at the end.
Emotional Support. Emotional support was not originally identified as a domain by investigators prior to review of the transcripts, yet emerged as the domain with the most comments in all patient groups. Emotional support was frequently mentioned by family members and health care workers as well. This domain includes components encompassing compassion, hope, and comfort.
It's like all the doctor knows is the medical end of it and doesn't really get involved personally at all. The nurse is the one who steps in, holds your hand and talks about his or her kids and wants to know about your kids and, you know, what's going on and how you're feeling.
(Patient with cancer)
Accessibility and Continuity. Accessibility and continuity was consistently 1 of the most frequently mentioned domains by both patients and family members. The major components of this domain include structural features of how medical care is organized (waiting times and interruptions) as well as features of the patient-physician interaction (giving the impression that the patient will not be abandoned prior to death). Two examples of quotes from patients coded in this domain are:
I have a vision of my doctor being there for me when I'm getting ready to die. And if he's not, I'm gonna curse his ass out, next time I see him, because I didn't come all this way with him to be left.
(Patient with AIDS)
I had a friend die a year ago. His doctor was accessible; he was there. He didn't become a friend of this person, but he certainly became an intimate part of this person's death.
(Patient with cancer)
A nurse also expressed an important perspective on the issue of continuity and abandonment demonstrating some overlap with the domain of emotional support:
There's that sense of abandonment that comes when there are fewer interventions being offered, fewer reasons to go to see the doctor, or less help to be able to go in to see the doctor. Those little things like phone calls become the line they still have to somebody, a physician who cares.
Competence. Competence was ranked, by frequency, in the upper half of all domains for each of the patient groups and for family members. In contrast, it ranked 6 of 12 for nurses and social workers and 11 of 12 for physicians. Components within competence not only focused on knowledge, skill, and use of referrals, but also on taking patients' symptoms seriously and knowing when to stop treatments. Examples of quotes from this domain are:
They're in the business of keeping you going and keeping you encouraged and trying this and trying that, and when it gets to that point where you're dying, you want them to stop. You want them to know when to stop.
(Patient with AIDS)
Our doctor could have benefited from having some hospice training and understanding, because she was giving us conflicting information. We know hospice was the expert in this case, but it caused a lot of dissension in my family.
Respect and Humility. The domain respect and humility encompassed components having to do with treating patients as equals, admitting one's own shortcomings, and not viewing death as a personal failure. While this domain was ranked in the lower half of domains for patients and families, it was the domain most commonly identified by physicians and the third most common for nurses and social workers.
Our doctors really distanced at the end, when it was pretty clear that things looked bad and when my mother actually [was dying] they just distanced. They got cold, they got mad, they lost their sense of humor, they lost their humanity, and so they acted to my mom in a far different way than they would have a week earlier when things were going better. It's like they just couldn't handle it.
If you think “it's just another IV drug abuser dying of something that they deserve,” you're going to be bad at helping them die or helping them through it… Whereas it's very different if you give value to their life and you respect the fact that they're a human being the same way the rest of us are.
Team Communication and Coordination. Team communication and coordination was identified more frequently by health care workers than by patients and families. This domain includes structural components (making sure someone is available to help when the physician is unavailable) as well as personal characteristics of the physician (respects and uses the expertise of nonphysician team members). Examples of quotes from this domain are:
My doctor worked with my mental health professional from another agency. They worked very well together. He's not an exclusivist; he doesn't see himself as the almighty doctor.
(Patient with AIDS)
I think the biggest frustration I've ever had in my whole 8 years is not being [part of] a team… I think that the hardest thing for me is when doctors, because of their pride or whatever, feel that they have to have complete control and negate any experience or knowledge you have and won't listen.
Patient Education. Components from patient education focused on the provision of information to patients and families in ways they could understand and that were meaningful to them in their lives. Prior to review of the transcripts, this domain was not separate from communication with patients but emerged as clearly distinct after transcript review. For example:
It is good to get a little technical. I don't like it when the doctor says, “Oh, your tests are good.” So what does that mean? I like him to say, “Your blood count is this and your hematocrit is that…” I have lung cancer, so I've had a jillion chest x-rays and the first doctor I had would show me them and put it in layman's terms… It gave me a good idea of what was happening.
(Patient with cancer)
Another thing that is really helpful is a physician who can assess where any given person is with their knowledge and take it from there with trying to explain what's going on. Make it very simplistic for those who need it and equally as sophisticated for those who have the ability or the background to handle the information. Tailoring information is really helpful—being able to be the teacher and explain things so that a group of family members can walk away hearing the same message.
Personalization. Similar to patient education, personalization was not identified as a domain by the investigators prior to reviewing transcripts. This domain was ranked 5 or lower of the 12 domains for all groups. Components of this domain focus on understanding and treating the patient as a unique individual.
When you're sent out to the specialty clinics, the quality of the personalization of care goes way downhill. They seem to be terribly busy, and it's in and out the door, and they treat you like a number. It's very depersonalizing and demoralizing.
(Patient with AIDS)
It appears that the people who are really speaking highly of their doctors and have a good relationship with them, I think they feel the doctor knows them, not just as a patient, but as a person. They also feel that they know the doctor, not just as a doctor, but as a person.
(Patient with COPD)
Pain and Symptom Management. Pain and symptom management incorporates treating pain and symptoms, including anxiety and depression, as well as incorporating patient preferences for alertness and educating patients and families about symptoms and treatments. This domain did not rank in the top half of domains for any of the groups. Nonetheless, some of the comments belied the importance of this domain.
The most important thing is to know how to treat pain. And when I talk about pain, I'm identifying shortness of breath as pain. To know how to treat pain aggressively with one who is dying, I think that it's probably, to me, priority number 1.
(Patient with COPD)
He was unable towards the end to verbalize… he would moan or he would nod his head… there were indications he was having pain… really having a lot of pain. I just didn't feel that his doctor was thorough enough…[The patient] would get mad, he'd say, “I'm in a lot of pain! Let's do something! Can you give me something? I'm really hurting here!”… It was very, very frustrating. We were all horrified that he had to suffer so much and it was unnecessary.
Inclusion and Recognition of the Family. Inclusion and recognition of the family was the fourth most commonly identified domain among family members and physicians. The three components identified within this domain focus on communication with and inclusion of the family.
My doctor involved my family in my care when I was in the doctor's office, he involved them in my care when I was in the hospital, all along the way. I think your family is a very important part of your care and your support system and your doctor has to keep them involved.
(Patient with AIDS)
[The doctor] was very inclusive of all the family members and the support system and describing what [the patient's] situation was, what were the options, what were some of the outcomes both good and bad. It always seemed to leave a positive note even though something dreadful was happening.
Attention to the Patient's Values. The domain concerning attention to patient's values encompassed respecting patients' beliefs about a range of topics, including religion and spirituality, alternative medicine and lifestyles. One component, not blaming or judgmental about lifestyle, included features such as sexual orientation, drug use, and cigarette smoking.
The doctors were willing to work with us… at least be open to and not condescending of our wish to try alternative types of health care… We both felt very strongly that allopathic medicine had certain limitations and the naturopathic approach can offer some benefits… I've heard some horror stories of some doctors being adamant that you maintain the medical treatment plan.
Any doctor I like is not judgmental. Any doctor that asks me if I smoke in a judgmental way, I don't want to deal with him, because I think he's got me in a slot.
(Patient with COPD)
Support of Patient Decision Making. The domain support of patient decision making was identified least commonly overall and ranked in the bottom half for all groups. Nonetheless, this domain was identified by every group. Components focus on making patients' informed wishes the center of decision making about medical care at the end of life.
My worst experience was a doctor who yelled at me, scared me to death. He suggested that I do something to slow the progression of my disease, and I told him that I didn't know if I wanted to and I would like to take some time to think about that, and he was extremely irate that I would question his judgment and authority.
(Patient with cancer)
It's beyond giving you all the information that you need to make a decision. They all seem to be pretty good at that. I would say it is just a sense that… this person really cares that we make the right decision for us. With a lot of them, it's sort of like a disclaimer, you know, they put out all the information and say “Well, it's your decision. Let me know.” But they should really be concerned that, whatever your decision… they've done everything they can to help you make the best decision for you.
(Patient with cancer)
We developed a conceptual model of physicians' skills at providing high-quality end of life care using the content and context of participants' comments to identify connections between domains (Fig. 1). This model groups the domains into five different categories and its structure is based on analyses of the transcripts. For example, the position and relationship of the categories of cognitive skills and affective skills at the top of the model are based on the qualitative importance that patients placed on these categories and on quotes that describe an interaction between these two categories. The following examples demonstrate this relationship between cognitive and affective skills resulting in the joining of these two boxes in the model:
Knowing that life is coming to an end… you're asking [doctors] to be superhuman. Beyond all the technical things that physicians are doing, to somehow encourage a patient to talk with them about their feelings… to set the stage to allow… some of these things to come out.
It's like having a country doctor who has the knowledge of a city doctor.
(Patient with AIDS)
My main attending medical oncologist was very competent, his reputation is good… In terms of his knowledge, his technical skills, he was very good. In terms of the emotional issues, he was horrible… Only twice in my course of treatment did I ever express emotional issues. One time I said, “I feel really frightened,” and he just got really irritable and he goes, “Well, we're doing the best we can.”
(Patient with cancer)
The three categories below this dyad of cognitive skills and affective skills that complete the circle encompass the major categories of (1) overall communication skills, (2) focusing on values that are important to patients, and (3) having systems of care that allow individual physicians to provide high-quality end of life care, and that also encourage a focus on what is important to individual patients. There were many examples where the interaction between the system and the physician seemed to be the foundation for physician skill at providing end of life care. In the first example below, this foundation exists, whereas in the second example, the view expressed is that the absence of effective coordination between systems of care and the individual doctor in many places leads to problems with the quality of care:
Our doctor had a triage nurse, that was her job, was really answering phone calls. She was really an intermediary, but she was wonderful. She was very compassionate, always got back to us by 5:00 the same day. That made a huge difference in how we felt about the care
I've always been impressed with Seattle, and all its organizations. I've tried to live in five or six states and I think Washington is the best state for someone with AIDS. Because of not just the physician, but because of all the other organizations around them. They all work as a team. You can have a great country doctor with a great education, but if you don't full circle that with all the other services, it doesn't matter how good the doctor is.
(Patient with AIDS)
We used focus groups of patients with AIDS, cancer, or COPD as well as focus groups of family members and health care workers to develop an understanding of the domains and specific components of physicians' skills at providing quality end of life care. The focus group methodology permitted identification of the overall domains of end of life care from these different perspectives as well as the major specific components in the words of the patients and families. The findings suggest that physicians' skills at providing end of life care fall into 12 domains. The 12 domains and 55 components pinpoint specific areas that physicians could target in their care of patients near the end of life. Further, we developed a conceptual model that provides a broader context in which these domains may be understood, taught, and evaluated.
Although several prior reports identified domains of quality end of life care, those reports either represented the perspectives of health care professionals and experts3,10,11 or used interviews with patients conducted for a different purpose that may have directed patients toward specific topics.12 Our study incorporated the perspectives of terminally ill patients and their family members in addition to health care professionals, and is the first report to provide a focus on the specific components of quality of end of life care provided by physicians. Although physicians are only one part of the medical system that provides care at the end of life, they play an integral role and have been found deficient in end of life care.3 These domains and specific components within the domains could be evaluated for targeted educational and research interventions to improve the quality of end of life care. Based on the findings of this study, such interventions should focus on areas of importance to patients such as communication and emotional support, skills that are not currently taught, emphasized, or supported in many medical schools and residency training programs.19
Patients participating in the construction of this understanding of physicians' skills had three specific diseases: cancer, COPD, and AIDS. Cancer and COPD represent the second and fourth most common causes of death in the United States, respectively,20 and AIDS is the most common cause of death for persons aged 15 to 44 years.21 The domains identified by patients with these three diseases showed remarkable similarity in that each of the 12 domains was identified in all disease groups. In addition, the most commonly referenced domains were similar across the disease groups. This suggests that our findings may be generalizable to other diseases and that efforts to evaluate the model, domains, and components and to develop educational and other interventional programs could include other diseases.
The conceptual model represents a tool for understanding and improving the quality of physicians' skills at providing end of life care. At the top of the model is the dyad of cognitive skills and affective skills, focusing physicians and physician-educators on the importance both components. Finding innovative ways to teach affective skills is one of the challenges facing medical education concerning end of life care.19 In addition, the model highlights communication skills and patient-centered values as key components in the overall model. Finally, the model and the words of patients and families remind us that we are unlikely to effect high-quality end of life care unless we can develop and support systems that provide for patient-centered care and train physicians to interact well with these systems.
This study has several limitations. All focus groups were conducted in one geographic region, and with the exception of the patients with AIDS, focus group members were predominantly white. There may be geographic or cultural differences important to understanding physicians' skills at providing high quality end of life care. Future studies should address the generalizability of this study to other areas and other cultural and ethnic groups. Furthermore, because we included patients with three specific chronic diseases, there may be important differences among patients with diseases not represented in this study or among patients at more advanced stages of illness in long-term care facilities. In addition, the number of utterances does not necessarily equal importance of a domain in focus groups, in part because of the possibility that a few participants could dominate the group. However, study methods were designed to avoid letting one or a few participants dominate discussions, and similarities across groups support the generalizability of findings. Finally, the domains, components, and model proposed in this study should be evaluated to assess their usefulness for improving quality of care.
The domains we identified encompass the major aspects of physicians' skills at providing high-quality end of life care from the perspectives of patients, their families, and health care workers and provide a framework for understanding, evaluating, and teaching these skills. The 55 components of these domains were identified in the language of the recipients of medical care, patients and families, and provide specific markers for measuring and improving the quality of end of life care provided by physicians. These findings could be used to develop educational programs for medical students and physicians, design interventions for health care settings, and develop an evaluation tool, with the goal of improving the quality of care that physicians provide to patients at the end of life. Further work is needed concerning the generalizability of domains and components and to evaluate the usefulness of the domains, components, and model as a framework for understanding and improving the quality of care provided by physicians at the end of life.
Project support was provided by the Open Society Institute Project on Death in America. The opinions expressed here are those of the authors and not necessarily those of the funding organization.